JP7646105B2 - OBJECT DETECTION SYSTEM AND OBJECT DETECTION METHOD - Google Patents

Description

This disclosure relates to object detection using lasers.

Some LiDAR measurements utilize the ToF principle.
LiDAR is an abbreviation for Light Detection And Ranging.
ToF is an abbreviation for Time of Flight.

Measurements using the ToF principle calculate distance based on the time it takes for laser light to be emitted and for the laser light (reflected light) to be reflected off the subject and received by the sensor.

LiDAR mainly uses near-infrared light with a wavelength of 903 nanometers or 905 nanometers. LiDAR with a wavelength of 1550 nanometers is also under development. The longer the wavelength, the longer the distance that can be measured.
In 900 nanometer LiDAR, silicon (Si) is used for the light receiving element.
In the 1550 nanometer LiDAR, indium gallium arsenide (InGaAs) is used for the light receiving element.

A spoofing attack using false reflected waves is known as an attack against an infrared sensor that measures the surrounding environment, such as LiDAR. Countermeasures against this spoofing attack include redundancy and sensor fusion.
However, there are known attacks against self-driving cars that deceive the object detection signal processing of sensor fusion using cameras and LiDAR.
Therefore, it is important to take measures against attacks using individual sensors, in addition to measures using sensor fusion with multiple sensors.

Known countermeasures to attacks on sensors include random modulation, the use of multiple wavelengths, and reducing the light receiving angle.
Randomly modulated FMCW LiDAR has been developed, but it is difficult to rotate and has a narrow measurement range. Multiple LiDARs are required to perform 360-degree measurements, making it unsuitable for small devices such as drones.
In the case of the adoption of multiple wavelengths, an attacker can carry out a low-cost attack by using multiple wavelengths.
Reducing the light receiving angle can reduce the room for an attack. However, it is costly to install multiple LiDARs. There are use cases where a cost-effective countermeasure against sensor attacks is required without increasing the number of LiDARs.

Patent document 1 discloses a countermeasure against deception attacks on sensors. This countermeasure involves preparing a deceptive signal, making an attacker believe that the deceptive signal is a true measurement signal, and then attacking the deceptive signal.

Japanese Patent Application Publication No. 9-281397

Conventional countermeasures against spoofing attacks require multiple LiDARs, which cannot be applied to small devices such as UAVs, and also involve financial and technical costs.
UAV is an abbreviation for Unmanned Aerial Vehicle.

No countermeasures have been proposed against attacks that deceive the sensor fusion signal processing.

The purpose of this disclosure is to prevent erroneous object detection caused by spoofing attacks.

The laser sensor of the present disclosure includes:
A main laser that irradiates laser light of a specific wavelength;
a dummy laser that irradiates a laser beam having a wavelength different from the specific wavelength;
a light receiving element having sensitivity to light of the specific wavelength;
Equipped with.

This disclosure makes it possible to prevent erroneous object detection caused by spoofing attacks.

FIG. 1 is a configuration diagram of an object detection system 100 according to a first embodiment. FIG. 2 is a configuration diagram of a laser sensor 110 according to the first embodiment. FIG. 2 is a configuration diagram of an attack detection device 120 according to the first embodiment. FIG. 2 is a configuration diagram of an object detection device 130 according to the first embodiment. FIG. 1 is a functional configuration diagram of an object detection system 100 according to a first embodiment. 4 is a flowchart of an object detection method according to the first embodiment. 4 is a flowchart of an object detection method according to the first embodiment.

In the embodiments and drawings, the same or corresponding elements are given the same symbols. Explanations of elements given the same symbols as the elements described above are omitted or simplified as appropriate. Arrows in the drawings primarily indicate the flow of signals, data, or processing.

Embodiment 1.
The object detection system 100 will be described with reference to FIGS.

***Configuration Description***
The configuration of an object detection system 100 will be described with reference to FIG.
The object detection system 100 is a system that detects a subject 101 using a laser sensor 110 .
The subject 101 is an object to be detected. In other words, the subject 101 is an object to be sensed.
The object detection system 100 includes a laser sensor 110, an attack detection device 120, and an object detection device 130.
Specifically, the laser sensor 110 is a LiDAR, which is an infrared sensor, and the attack detection device 120 and the object detection device 130 are computers.

The configuration of the laser sensor 110 will be described with reference to FIG.
The laser sensor 110 includes three types of lasers (111, 112, 113).
The first main laser 111 and the second main laser 112 are lasers for measurement, and irradiate laser light of different wavelengths.
The first main laser 111 radiates a laser beam having a first wavelength. Specifically, the first main laser 111 radiates a laser beam having a wavelength of about 900 nanometers.
The second main laser 112 radiates a laser beam having a second wavelength. Specifically, the second main laser 112 radiates a laser beam having a wavelength of about 1550 nanometers.
The dummy laser 113 irradiates a laser beam having a third wavelength. The third wavelength is a wavelength between the first wavelength and the second wavelength. Specifically, the dummy laser 113 irradiates a laser beam having a wavelength of about 1000 nanometers.

The laser sensor 110 includes two types of light receiving elements (114, 115).
The first light receiving element 114 is a light receiving element that has sensitivity to light of a first wavelength. Specifically, the first light receiving element 114 is a light receiving element that uses a silicon (Si) semiconductor.
The second light receiving element 115 is a light receiving element having sensitivity to light of the second wavelength. Specifically, the second light receiving element 115 is a light receiving element using a compound semiconductor of indium gallium arsenide (InGaAs).

The laser sensor 110 has a mechanism for rotating three types of lasers (111, 112, 113) and two types of light receiving elements (114, 115) as a whole in order to perform measurements over a wide surrounding range.
For example, the laser sensor 110 has a rotation mechanism for performing measurements around 360 degrees. The rotation mechanism rotates the three lasers and the two light receiving elements in their entirety through 360 degrees.

The configuration of the attack detection device 120 will be described with reference to FIG.
The attack detection device 120 includes hardware called a processing circuit 129 .
The attack detection device 120 includes an element called an attack detection unit 121 .
The processing circuit 129 is hardware that realizes the attack detection unit 121 .
The processing circuitry 129 may be dedicated hardware or may be a processor that executes a program stored in a memory.
Dedicated hardware may be, for example, a single circuit, multiple circuits, a programmed processor, parallel programmed processors, an ASIC, an FPGA, or a combination of these.
ASIC is an abbreviation for Application Specific Integrated Circuit.
FPGA is an abbreviation for Field Programmable Gate Array.

The configuration of the object detection device 130 will be described with reference to FIG.
The object detection device 130 includes hardware called a processing circuit 139 .
The object detection device 130 includes an element called an object detection unit 131 .
The processing circuit 139 is hardware that realizes the object detection unit 131 .
The processing circuitry 139 may be dedicated hardware or may be a processor that executes a program stored in a memory.

FIG. 5 illustrates the relationships between elements in object detection system 100.
The three types of lasers (111, 112, 113) emit laser light, but the two types of main lasers (111, 112) emit laser light alternately.
Two types of light receiving elements (114, 115) receive (detect) the laser light reflected by the subject 101 and output a light receiving signal.
Receiving light means receiving light and converting the received light into an electrical signal.
The detection means regarding the electrical signal as point cloud data. When the amount of light is greater than a threshold, the electrical signal is regarded as point cloud data. When the electrical signal is not regarded as point cloud data, the electrical signal is discarded as unnecessary data.
The received light signal is an electrical signal that is received and detected and represents the laser light.
The attack detection unit 121 receives the light reception signal, detects an attack by the attack device 109, and outputs a light reception signal indicating that there is no attack.
The object detection unit 131 processes the unattacked light reception signal to detect the subject 101 .

The attacking device 109 is a device that performs a spoofing attack.
The attack device 109 detects the laser light emitted from the three types of lasers and emits jamming laser light.

*** Operation Description ***
The procedure of operation of the object detection system 100 corresponds to an object detection method.

The object detection method will be described with reference to FIGS.
The operation of the laser sensor 110 in the object detection method will be described with reference to FIG.
In step S101, the first main laser 111 and the second main laser 112 alternately irradiate laser light. In addition, the dummy laser 113 continues to irradiate laser light.

The first main laser 111 emits a laser beam having a first wavelength. The laser beam having the first wavelength is referred to as a first laser beam.
The second main laser 112 irradiates a laser beam having a second wavelength. The laser beam having the second wavelength is referred to as a second laser beam.
The dummy laser 113 irradiates a laser beam having a third wavelength. The laser beam having the third wavelength is referred to as a dummy laser beam.

Each of the first laser light, the second laser light, and the dummy laser light hits the subject 101 and is reflected.

When the attack device 109 detects the first laser light, the second laser light, or the third laser light, it emits a jamming laser light. The jamming laser light is referred to as the attack laser light.

In step S102, the first light receiving element 114 and the second light receiving element 115 receive laser light.

The first light receiving element 114 detects the first laser light. When the wavelength of the attack laser light is close to the first wavelength, the first light receiving element 114 detects the first laser light and the attack laser light. On the other hand, the second laser light and the dummy laser light are not detected by the first light receiving element 114.
The second light receiving element 115 detects the second laser light. If the wavelength of the attack laser light is close to the second wavelength, the second light receiving element 115 detects the second laser light and the attack laser light. On the other hand, the first laser light and the dummy laser light are not detected by the second light receiving element 115.

In step S103, the first light receiving element 114 and the second light receiving element 115 output light receiving signals.
The light receiving signal output from the first light receiving element 114 is referred to as a first light receiving signal.
The light receiving signal output from the second light receiving element 115 is referred to as a second light receiving signal.

The operations of the attack detection device 120 and the object detection device 130 in the object detection method will be described with reference to FIG.
In step S<b>111 , the attack detection unit 121 receives a light receiving signal from the first light receiving element 114 and the second light receiving element 115 .

In step S112, the attack detection unit 121 determines the S/N ratio of the received optical signal. The S/N ratio means the signal-to-noise ratio.

The signal-to-noise ratio of the received signal is determined as follows.
First, the attack detection unit 121 calculates the S/N ratio of the received light signal.
Then, the attack detection unit 121 compares the S/N ratio with a threshold value.
If the S/N ratio is equal to or greater than the threshold, the attack detection unit 121 determines that the S/N ratio is good.
If the S/N ratio is less than the threshold, the attack detection unit 121 determines that the S/N ratio is poor.

If the S/N ratio is good, the process proceeds to step S113.
If the S/N ratio is poor, the received light signal is considered to be a dummy laser light signal. Therefore, the received light signal is removed as an unnecessary point cloud. In other words, the received signal is removed as an electrical signal that does not become point cloud data. Then, the process ends without object detection.

In step S113, the attack detection unit 121 determines whether the first light receiving signal from the first light receiving element 114 and the second light receiving signal from the second light receiving element 115 are received simultaneously.
If the first and second light receiving signals are not received simultaneously, the process proceeds to step S114.
When the first and second light receiving signals are received simultaneously, one of the light receiving signals is considered to be an attack signal. Therefore, the light receiving signal is processed as a false reflected wave. This means that although the light receiving signal is valid point cloud data, it is an attack signal, and therefore the point cloud is removed so as not to affect the subsequent application. The subsequent application means, for example, processing by the processing circuit 139. Then, the processing ends.

In step S<b>114 , the attack detection unit 121 determines whether or not it has received both the first light receiving signal from the first light receiving element 114 and the second light receiving signal from the second light receiving element 115 .
If both the first light receiving signal and the second light receiving signal are received, the attack detection unit 121 outputs both the first light receiving signal and the second light receiving signal. Then, the process proceeds to step S121.
If at least one of the first and second light receiving signals is not received, the received light receiving signal is considered to be an attack signal, and the process ends without performing object detection.

In step S121, the object detection unit 131 receives each of the first light receiving signal and the second light receiving signal.
The object detection unit 131 then performs object detection using at least one of the first light reception signal and the second light reception signal, thereby detecting the subject 101.

The object detection process involves the following steps:
The object detection unit 131 calculates the relative distance at the measurement time based on the time of flight of the laser light.
The relative distance is the distance from the laser sensor 110 to the subject 101, and is calculated by multiplying the flight time by the speed of light.
The time of flight is the time from when the laser light is emitted to when the laser light is received (detected), and is measured based on the number of clocks of an internal circuit, for example.
The measurement time is the time when the laser light is emitted or the time when the laser light is received (detected).

After step S121, the processing ends.

Steps S101 to S121 are executed repeatedly.

***Advantages of First Embodiment***
The first embodiment makes it possible to take measures against deception attacks on sensors and advanced attacks that deceive sensor fusion.
According to the first embodiment, it is possible to significantly reduce the deception attack capability of an attacker. The attacker can provide a new mechanism for removing signals with poor S/N ratios. However, an attack that overcomes sensor fusion requires the insertion of point clouds in specific locations, and is difficult to achieve if the degree of freedom of the attack is reduced. In addition, since the attacker needs to add a new mechanism in addition to the attack mechanism, it is expected that the attack cost of the attacker will increase.

Summary of the First Embodiment
In the following description, elements corresponding to those in the first embodiment are placed in parentheses, and the reference numerals of the elements in the first embodiment are written in parentheses.
The infrared sensor (110) has a unique configuration and detects signals processed by the sensor.
The laser (111) of about 900 nm and the laser (112) of about 1550 nm are alternately irradiated. However, when the laser light of about 900 nm is irradiated, the laser light of about 1550 nm is received but is not processed as a point cloud. Similarly, when the laser light of about 1550 nm is irradiated, the laser light of about 900 nm is received but is not processed as a point cloud.
Laser light of about 1000 nm is irradiated but not received, and is treated as dummy light. Although the Si light receiving element (114) and the InGaAs light receiving element (115) react to light of about 1000 nm wavelength, the signal S/N ratio is poor. Therefore, the signal of light of about 1000 nm wavelength is removed in the processing stage of forming a point cloud.

Signal processing for attack detection distinguishes between real reflected waves and false reflected waves (signals generated by an attacker).
First, if light of the same wavelength arrives from the same position in alternating measurements, the arriving light is regarded as a false reflected wave. This makes it possible to prevent laser irradiation attacks using either the wavelength of about 900 nm or about 1550 nm.
Next, when both the light of the approximately 900 nm laser (111) and the light of the approximately 1550 nm laser (112) arrive at the same time, the arriving light is regarded as a false reflected wave. This is because the light of approximately 900 nm and the light of approximately 1550 nm are alternately emitted, so that only one of the lights should be detected at a time. Even if an attacker prepares a laser irradiation attack device (109) for both wavelengths, the attacker's light receiving mechanism reacts to the dummy light of approximately 1000 nm, and the laser irradiation attack device emits laser light of both wavelengths. This makes it possible to detect the attack. This is because the attacker does not have a signal processing mechanism to remove signals with poor S/N ratios.

The first embodiment is characterized by using dummy light, which has low measurement sensitivity of the light receiving elements (114, 115).
Instead of using completely invisible light as dummy light, light that can be measured is used as dummy light, which can induce attackers to attack and eliminate false signals.

***Additional Notes to the First Embodiment***
The light receiving signal output from the attack detection unit 121 is used, for example, for object detection based on a collection of point clouds and for mapping the surrounding environment.

In an attack that overcomes sensor fusion, it is not possible to simply shine light on the target; it is necessary to shine light on the target in a way that causes a point cloud to appear in a specific location and with a specific pattern. For example, a specific location would be behind an obstruction such as a car. For example, a specific pattern would be to shine light on the target so that the point cloud appears as a cone. It is difficult to shine light on a moving object, so an elaborate attack requires a high degree of freedom in the attack (few conditions for the attack to be successful). Therefore, if the degree of freedom in the attack is reduced, it becomes difficult to implement.

Embodiment 1 is an example of a preferred embodiment and is not intended to limit the technical scope of the present disclosure. Embodiment 1 may be implemented in part or in combination with other embodiments.

The laser sensor 110 does not necessarily have to include one of the first main laser 111 and the second main laser 112 .
The attack detection device 120 and the object detection device 130 may be realized in a single device.

The procedures explained using the flowcharts etc. may be modified as appropriate.
The order of steps S112, S113, and S114 may be interchanged.

The "part" of each element of the attack detection device 120 and the object detection device 130 may also be read as "processing," "step," "circuit," or "circuitry."

The program can be recorded (stored) in a computer-readable manner on a non-volatile recording medium such as an optical disk or flash memory.

100 Object detection system, 101 Subject, 109 Attack device, 110 Laser sensor, 111 First main laser, 112 Second main laser, 113 Dummy laser, 114 First light receiving element, 115 Second light receiving element, 120 Attack detection device, 121 Attack detection unit, 129 Processing circuit, 130 Object detection device, 131 Object detection unit, 139 Processing circuit.

Claims (2)

a first main laser that irradiates a laser beam having a first wavelength, and a second main laser that irradiates a laser beam having a second wavelength,
the first main laser and the second main laser alternately irradiate the laser light,
a dummy laser that irradiates a laser beam having a third wavelength;
a first light receiving element that is sensitive to light of the first wavelength, detects a laser beam, and outputs a first light receiving signal that is a signal of the detected laser beam;
a second light receiving element that is sensitive to light of the second wavelength, detects the laser light, and outputs a second light receiving signal that is a signal of the detected laser light;
an attack detection unit that receives the output first light receiving signal and the output second light receiving signal, determines a signal-to-noise ratio of the received first light receiving signal and the received second light receiving signal, determines whether the first light receiving signal and the second light receiving signal have been received simultaneously , and removes the first light receiving signal and the second light receiving signal when it is determined that the first light receiving signal and the second light receiving signal have been received simultaneously ;
an object detection unit that performs object detection using at least one of the first light receiving signal determined to have a good signal-to-noise ratio and the second light receiving signal determined to have a good signal-to-noise ratio when it is not determined that the first light receiving signal and the second light receiving signal are received simultaneously;
An object detection system comprising: Alternately irradiating a laser beam having a first wavelength and a laser beam having a second wavelength;
Irradiating the laser light having a third wavelength;
a first light receiving signal outputted from the first sensor which is sensitive to light of the first wavelength and detects the laser light;
a second light receiving signal outputted from the second light receiving element, which is sensitive to light of the second wavelength and detects the laser light;
receiving the output first light receiving signal and the output second light receiving signal, determining a signal-to-noise ratio of the received first light receiving signal and the received second light receiving signal, and determining whether the first light receiving signal and the second light receiving signal are received simultaneously;
removing the first light receiving signal and the second light receiving signal when it is determined that the first light receiving signal and the second light receiving signal are received simultaneously;
An object detection method for detecting an object using at least one of a first light receiving signal determined to have a good signal-to-noise ratio and a second light receiving signal determined to have a good signal-to-noise ratio when it is not determined that the first light receiving signal and the second light receiving signal are received simultaneously.

Images